CN114743112B - Big data-based seed production corn field identification method and device and electronic equipment - Google Patents
Big data-based seed production corn field identification method and device and electronic equipment Download PDFInfo
- Publication number
- CN114743112B CN114743112B CN202210407010.1A CN202210407010A CN114743112B CN 114743112 B CN114743112 B CN 114743112B CN 202210407010 A CN202210407010 A CN 202210407010A CN 114743112 B CN114743112 B CN 114743112B
- Authority
- CN
- China
- Prior art keywords
- data
- target
- field
- period
- corn
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method and a device for identifying a seed production corn field based on big data and electronic equipment, wherein the method comprises the following steps: determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area; based on the spectral texture characteristics corresponding to the corn field blocks, a first area and a second area are identified from the corn field blocks, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified through the spectral texture characteristics; judging whether each field is a seed production corn field or not based on the target terminal signaling data corresponding to each field in the second area; the target terminal signaling data is used for indicating the farming flow information corresponding to each field. The method for identifying the seed production corn fields based on the big data can extract the farming flow information corresponding to each field block in the second area based on the signaling data of the target terminal, is beneficial to improving the identification precision of the seed production corn fields, and can also quickly and effectively identify the seed production corn fields.
Description
Technical Field
The invention relates to the technical field of crop type identification, in particular to a method and a device for identifying a seed production corn field based on big data and electronic equipment.
Background
The corn is taken as main grain crop in China, the production area of the seed production corn is monitored rapidly and accurately, and the urgent requirement of ensuring the agricultural seed supply safety and strengthening the seed production supervision is met. The traditional crop area acquisition is generally obtained by summarizing all levels of seed management departments, and the method is greatly influenced by human factors, has low efficiency and low speed, and is not suitable for monitoring and illegal seed production monitoring of large-area seed production corns. The method can obtain the area and space distribution information of crop planting in time and objectively, and is a necessary choice for realizing accurate supervision of seed production corn production.
Researchers at home and abroad have developed method exploration in the aspect of identifying the corn seeds. The existing method is mainly based on spectral characteristics, texture structures and other image information of the seed corn or combines the physical characteristics of crops, the ground science characteristics of background environment and the like of the seed corn, and the supervision classification method is used for directly identifying the seed corn. The above method has the following limitations:
from the aspect of data characteristics, researchers mainly depend on various optical remote sensing satellite data, and combine spectral characteristics such as vegetation indexes and the like of seed corn and texture characteristics such as gray level co-occurrence matrix and the like to realize regional seed corn field identification. However, the method is based on optical remote sensing satellite data, the feasibility is determined by the spatial resolution of the optical remote sensing satellite data to a great extent, and meanwhile, the imaging quality of the optical remote sensing satellite data directly influences the effectiveness of spectral features and texture features of vegetation canopy, so that the identifiability of seed corn is influenced, and spectral response failure phenomenon can occur to corn fields with high canopy closure. In order to solve the problems, researchers aim to enhance the expression of spectral features of crop canopy and weaken the defects by increasing the number of vegetation indexes and improving a texture feature mining mode, but the mode does not additionally increase ground object optical information basically, and has limited effect on improving the identification capability of seed corn. The learner establishes a vegetation index and other spectrum feature sets under multiple time phases based on different time intervals, so that spectrum features of the seed corn in different breeding periods are effectively increased, but the mode also greatly improves the acquisition requirement and the identification cost of the high-spatial resolution remote sensing image, and has limitations in large-area popularization and application of the method.
From the aspect of model algorithm, researchers mostly use a supervision classification mode, and use machine learning algorithms such as a support vector machine, a random forest and the like to develop fine recognition of the seed corn and evaluate the utility of various classifiers. However, the effectiveness of each supervised classifier is greatly influenced by factors such as the number and distribution of training sample points, and human factors applied to sampling of the sample points can cause the possibility of differential performance of the same classification algorithm in different researches. Previous studies have shown that adequate, high quality training samples are a prerequisite for the effectiveness of supervised classification.
In summary, in the aspect of model construction, the robustness of the existing method is greatly influenced by the number of sample points and the quality, and the existing method has limitations in terms of effectiveness, economy and applicability in large-area popularization and application.
Disclosure of Invention
The invention provides a method and a device for identifying a seed production corn field based on big data and electronic equipment, which are used for solving the technical problems of low identification precision and low identification efficiency of the seed production corn field.
The invention provides a seed production corn field identification method based on big data, which comprises the following steps:
determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
Identifying a first area and a second area from the corn field based on the spectral texture features corresponding to the corn field, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified by the spectral texture features;
judging whether each field is a seed production corn field or not based on target terminal signaling data corresponding to each field in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field.
In some embodiments, the determining the corn field based on the target sample point data in the target region and the target remote sensing data corresponding to the target region includes:
classifying the ground object types included in the target remote sensing data based on an unsupervised classification algorithm to obtain an optimal classification result of the target region;
and determining the corn field block based on the target sample point data and the optimal classification result.
In some embodiments, the determining whether each field is a seed corn field based on the target terminal signaling data corresponding to each field in the second area includes:
extracting position information of a terminal and time information corresponding to the position information based on the target terminal signaling data;
And determining that the at least one field is a seed corn field when the position information is at least one field in the second area and the time information is a spike period.
In some embodiments, the time information includes: the total residence time of the spike period in a single day, the average residence time of the seeding period in a day and the average residence time of the growing period in a day;
the location information includes: the number of single-day terminals in the spike period, the number of average terminals in the seeding period and the number of average terminals in the growing period;
and determining that the at least one field is a seed corn field before the position information is at least one field in the second area and the time information is a spike period, wherein the method further comprises:
determining that the position information is at least one field in the second area and the time information is a spike period when the position information and the time information meet a target condition;
the target conditions include:
the spike period single day residence time is within a time threshold range, and the time threshold range is determined based on the sowing period daily average residence time;
the number of the single-day terminals in the spike period is in a quantity threshold range, and the quantity threshold range is determined based on the number of the daily terminals in the seeding period;
The total residence time of the spike period in a single day is longer than the residence time of the spike period in a common day in the growing period;
the number of the single-day terminals in the spike period is larger than the number of the average terminals in the growing period.
In some embodiments, before the determining the corn field based on the target sample point data in the target region and the target remote sensing data corresponding to the target region, the method further comprises:
respectively acquiring initial sample point data, initial remote sensing data and initial terminal signaling data in the target area;
vectorizing the initial sample point data and the initial terminal signaling data to obtain the target sample point data and the target terminal signaling data;
preprocessing the initial remote sensing data to obtain target remote sensing data;
wherein the preprocessing operation includes at least one of: data fusion, orthographic correction, spatial registration, projection conversion, and data cropping.
In some embodiments, prior to said determining said corn field block based on said target spot data and said optimal classification result, further comprising:
carrying out vectorization processing on the binarization grid image corresponding to the target area to obtain an initial corn field range vector;
And superposing the initial corn field range vector and the high-resolution satellite remote sensing data in the target remote sensing data, and correcting the superposed data to obtain the target corn field range vector.
The invention also provides a seed production corn field identification device based on big data, which comprises:
the first determining module is used for determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
the identification module is used for identifying a first area and a second area from the corn field block based on the spectral texture characteristics corresponding to the corn field block, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified through the spectral texture characteristics;
the judging module is used for judging whether each field block is a seed production corn field or not based on the target terminal signaling data corresponding to each field block in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field.
The invention also provides electronic equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the seed production corn field identification method based on any one of the above big data when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method of producing a seed corn field identification based on big data as described in any of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements the big data based method of producing a seed corn field identification as described in any of the above.
According to the big data-based seed production corn field identification method, the big data-based seed production corn field identification device and the big data-based seed production corn field identification electronic device, the difference of the big field corn and the seed production corn in the farming flow is utilized, the target terminal signaling data used for representing the farming flow information is introduced into the fine remote sensing classification of the corn field, and the supplement discrimination of the seed production corn is developed; the limitation of the remote sensing identification of the seed corn based on the optical characteristics can be overcome by introducing the agricultural process characteristics based on the traditional optical remote sensing characteristics, and the identification precision of the seed corn field based on the supplementary identification of the seed corn based on the agricultural process characteristics is beneficial to improvement.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for identifying a seed production corn field based on big data;
FIG. 2 is a schematic diagram of spectral textures of a corn canopy of a seed production corn field identification method based on big data provided by the invention;
FIG. 3 is a general technical flow chart of a method for identifying a seed production corn field based on big data provided by the invention;
FIG. 4 is a high-resolution No. 2 image schematic diagram of a research area of the big data-based seed production corn field identification method;
FIG. 5 is a schematic diagram of classification results of corn fields in a research area by applying the big data-based identification method of the seed production corn fields;
FIG. 6 is a schematic diagram of the identification result of the research area seed production corn field by applying the big data based seed production corn field identification method provided by the invention;
FIG. 7 is a schematic structural diagram of a big data based seed production corn field identification device provided by the invention;
fig. 8 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The general idea of the embodiment of the invention is as follows:
firstly, distinguishing the types of ground objects in a target area based on target remote sensing data, and determining corn field blocks; then, the texture characteristic difference of a large Tian Yumi field and a seed production corn field is utilized to subdivide corn field blocks, and the seed production corn field is identified; and finally, carrying out supplementary discrimination on the corn field blocks which are not discriminated based on the terminal signaling data by utilizing the difference of the crop flow of the field corn and the seed production corn in the ear period, and obtaining all seed production corn fields in the target area.
It can be understood that before the identification of the seed production corn field based on big data, the field investigation is performed on the corn field in the research area, specifically, the field investigation is performed on the differences of the seed production corn in the research area and the field corn in the aspects of planting mode, farming flow, time node and the like. In addition, field corn field sample point data and seed production corn field sample point data are collected through a field investigation method and are used for corn field remote sensing extraction and precision evaluation of seed production corn field identification based on the large data.
1. Determining planting mode
And determining the planting mode of the corn field for seed production in the research area by a field investigation method. The seed production corn is a field for preparing a large amount of corn hybrid seeds, and the planting modes are generally divided into a starry planting method, a row ratio planting method and a starry planting method.
Generally, in a corn field for seed production, a mode of planting a male parent plant and a female parent plant according to a certain line ratio is called a line ratio planting method; the planting mode of dispersing the male parent plants in the female parent plants is called a starry planting method; on the basis of the row ratio planting method, the planting mode of dibbling the male parent plants in the female parent rows is called a row ratio plus star planting method.
For field corns, the male parent plants and the female parent plants are not distinguished, and the problem of planting row ratio is not generally related to the aspect of planting modes. The invention is suitable for a row ratio planting method and a row ratio and star-filling planting method.
2. Determining a farming flow
And determining the farming flow of the corn field for seed production in the research area by a field investigation method.
The difference between seed corn and field corn in the agronomic flow is mainly characterized in the spike period.
In the ear period, the corn seeds need to be organized to enter the corn fields for farming for a plurality of times: firstly, after the seed production corn is subjected to emasculation, the stamens of a female parent plant needs to be removed in time, female parent emasculation is performed for short, and the selfing of the pistil and stamens of the female parent of the seed production corn is avoided; secondly, water and fertilizer are required to be supplemented to the seed production corn before and after pollination, thirdly, the male parent plant is required to be cut after pollination of the seed production corn. For field corn, this is not generally done during the ear period.
3. Determining a time node
And determining the time node of the crop of the seed production corn field in the research area by a field investigation method. The method specifically comprises the steps of determining the approximate starting time of the sowing period, the growing period and the spike period of the corn, and facilitating the subsequent image acquisition, the signaling data acquisition and processing and the like.
The sowing period is the period in which the seeds are buried in the soil. The growing period is the collective term of the growing stage between the seeding period and the spike period, and comprises the periods of emergence, long leaves, jointing and the like. The spike period is the growth period from the time of jointing to the time of male pulling.
4. Sampling point
The corn sampling points in the research area are collected by using a handheld global positioning system receiver through a field investigation method and are used for corn field remote sensing extraction and seed production corn field identification precision evaluation based on big data. The types of spots include field corn spots and seed corn spots.
Various sample points need to be collected in standard and standard seed production corn fields, the sample point collection should not be too aggregated, and the sample points should be scattered in the whole research area in space so as to ensure that the sample points have representativeness and representativeness.
Fig. 1 is a schematic flow chart of a method for identifying a corn field for seed production based on big data. Referring to fig. 1, an embodiment of the present invention provides a method for identifying a corn field for seed production based on big data, which may include: step 110, step 120 and step 130.
In step 110, the target area may be an area where seed corn field identification based on big data is desired. The target remote sensing data can be medium-high resolution multispectral satellite remote sensing data, medium-high resolution hyperspectral satellite remote sensing data or high resolution satellite remote sensing data.
The method comprises the steps of roughly classifying land object types included in satellite remote sensing images by utilizing the advantages of a plurality of vegetation red edge wave bands of target remote sensing data and target sample point data acquired in the field to distinguish corn from other land object types, so that corn fields planted with corn can be obtained.
In some embodiments, prior to step 110, further comprising:
respectively acquiring initial sample point data, initial remote sensing data and initial terminal signaling data in a target area;
vectorizing the initial sample point data and the initial terminal signaling data to obtain target sample point data and target terminal signaling data;
preprocessing the initial remote sensing data to obtain target remote sensing data;
wherein the preprocessing operation includes at least one of: data fusion, orthographic correction, spatial registration, projection conversion, and data cropping.
The initial sample data may be sample data actually collected in the target area.
The sampling point data are selected according to the principle of random and uniform distribution, and the number of the selected points is determined according to the size of the land block and the uniformity of the field growth vigor. The number of the fetching points should be properly increased when the field is large and the growth situation is uneven, otherwise, the number of the fetching points can be properly reduced. The spot data may be selected based on the region of the corn field where the seed corn or field corn is located, and is not particularly limited herein.
The initial remote sensing data is satellite remote sensing data of the ear period (after the male parent is cut down) of the seed production corn in the target area. The satellite remote sensing data includes: middle-high resolution multispectral satellite remote sensing data, middle-high resolution hyperspectral satellite remote sensing data or high resolution satellite remote sensing data.
According to the embodiment of the invention, the satellite remote sensing image with the target area spatial resolution of 1 meter is obtained through the geographic information platform and is used as a reference base map for multi-source image spatial registration. In addition, the related vector data such as the boundary of the target area is obtained and is used for cutting the remote sensing satellite raster data and the like.
It should be noted that the spectrum resolution is in the range of λ/10, and the remote sensing information with the spectrum resolution of λ/100 is called hyperspectrum. The substantial difference between multispectral and hyperspectral is: the hyperspectral band is more, can reach hundreds, and the band is narrower. The multispectral is relatively few in wave bands, and only a few wave bands exist in the visible light and near infrared spectral regions.
The initial terminal signaling data are terminal signaling data acquired in the corn sowing period, the growing period and the ear period in the target area.
The signaling data can be recorded as long as the terminal user uses the actions of switching on/off, talking, short message, position updating, base station switching and the like. Through the information exchange between the terminal user and the base station, the space position of the user can be determined, and the space-time track of the people stream can be recorded relatively accurately.
The terminal signaling data has the following characteristics: firstly, the space-time record with large sample, wide coverage range and high user holding rate can better reflect the people stream behavior; secondly, anonymous data has good safety, does not have any personal attribute information, and does not relate to personal privacy; thirdly, the involuntary data is not voluntarily used, and the user passively provides information and cannot intervene in the investigation result; and fourthly, the dynamic real-time performance and the continuity are realized, the space positions of the mobile phone users at different time points in the continuous time section can be accurately reflected, and the possibility is provided for describing the position flow track of the people in the area.
After the initial sample point data, the initial remote sensing data and the initial terminal signaling data are acquired, the data are preprocessed, and three aspects are mainly described:
1. The preprocessing of the initial remote sensing data can be carried out on a remote sensing image processing platform, and the data preprocessing operations such as data fusion, orthographic correction, spatial registration, projection conversion, data cutting and the like are respectively carried out on the multi-source remote sensing data.
The data fusion is a technology of spatially registering image data of the same region acquired by different types of sensors, and then organically combining the advantages or complementarity of each image by adopting a certain algorithm to generate a new image.
The space registration is a process of registering vector data and transforming the coordinate position of raster data by adopting mathematical transformation methods such as translation, scaling, rotation and the like.
Orthographic correction is a process of resampling an image into an orthographic image by selecting some ground control points on an original image and simultaneously performing inclination correction and projection difference correction on the image by utilizing digital elevation model data in the image range.
Projection conversion is used to convert different projection coordinate system data.
Data clipping is the process of clipping raster data.
2. And carrying out preprocessing on the initial terminal signaling data, screening and mining relevant parameters such as space coordinates and time nodes which can represent the position flow track of the user in the terminal signaling data, and carrying out vectorization processing on the terminal signaling data parameters representing the position flow track of the terminal user.
3. And carrying out preprocessing of initial sample data, and carrying out vectorization processing on the sample data acquired in the field based on the spatial position information.
And obtaining target sample point data, target remote sensing data and target terminal signaling data through a data preprocessing step. And the multi-source data such as the target sampling point data, the target remote sensing data, the target terminal signaling data and the like are distributed into the same geographic reference system.
According to the large data-based seed production corn field identification method, the initial sample point data, the initial remote sensing data and the initial terminal signaling data are preprocessed to obtain the multi-source characteristics such as the target sample point data, the target remote sensing data and the target terminal signaling data, so that the rapid and effective identification of the seed production corn field can be realized.
Based on the difference of spectral texture characteristics of the corn seed in the spike period and the corn canopy in the field, the first area and the second area can be identified by adopting a visual interpretation method.
Due to the influences of factors such as inconsistent farming progress of the seed production corn fields, poor imaging quality of remote sensing satellites and the like, the satellite remote sensing images can have the phenomenon that the spectral texture features of individual field corn blocks are fuzzy or the spectral texture features of the local seed production corn fields are not obvious, so that the seed production corn and the field corn cannot be distinguished only by virtue of the Tian Gao resolution spectral texture features of the corn and the visual interpretation method.
Therefore, the seed production corn field determined by the spectral texture features is determined as a first area, the corn field which cannot be determined by the spectral texture features is determined as a second area, and the second area is further complemented and identified.
In actual practice, a third region can be identified by spectral texture features, the third region being field corn.
It is understood that the corn field identified based on the target spot data and the target remote sensing data is a complete corn field comprising a plurality of corn fields. The first region, the second region, and the third region may each comprise one or more corn fields.
In some embodiments, the spectral texture features include streak features and homogeneity features;
identifying a corn field block exhibiting a striped characteristic as a seed corn field, i.e., a first region;
Determining a corn field corresponding to the features other than the streak feature and the homogeneity feature as a second region;
the corn field pieces exhibiting homogeneous characteristics are identified as field corn fields, i.e., the third region.
In actual implementation, female parent emasculation treatment and seed production corn male parent deforestation treatment are carried out in the seed production corn ear period, and spectrum textures of the seed production corn male parent deforestation treatment show stripe characteristics on a medium-high resolution hyperspectral (or multispectral) satellite remote sensing image. The spectrum texture of the corn ear period in the field is relatively homogeneous because the female parent emasculation and the male parent deforestation are not carried out. Examples of canopy spectral textures of spike stage seed corn and field corn are shown in fig. 2.
Therefore, the corn field blocks with the stripe characteristics on the satellite remote sensing image can be identified as seed corn fields, the corn field blocks with the homogeneous characteristics on the satellite remote sensing image can be identified as field corn fields, and the corn field blocks with the stripe characteristics and the characteristics other than the homogeneous characteristics on the satellite remote sensing image are identified as second areas, wherein the second areas are corn fields needing further identification.
According to the big data-based seed production corn field identification method, the spectral texture features are identified, so that the corn field blocks are judged in steps, the corn field blocks which cannot be identified in planting types through the spectral texture features can be subjected to supplementary judgment, and the accuracy of seed production corn field identification is improved.
And 130, judging whether each field is a seed corn field or not based on target terminal signaling data corresponding to each field in the second area, wherein the target terminal signaling data is used for indicating farming flow information corresponding to each field.
In this step, based on the space-time track of the terminal user recorded by the signaling data of the target terminal, the farming flow information corresponding to each corn field in the second area can be determined, and then the second area with the undetermined planting type is subjected to secondary supplementary recognition.
In actual implementation, the target terminal signaling data corresponding to each corn field block in the second area can be extracted one by one, then each corn field block in the second area is identified, and whether each corn field block is a seed production corn field or not is judged.
It is understood that a certain corn field block may be determined to be a seed corn field in the event that the crop flow information corresponding to the certain corn field block corresponds to the crop flow characteristics of the seed corn field.
Under the condition that the farming flow information corresponding to a certain corn field block accords with the farming flow characteristics of the field corn, the certain corn field block can be determined to be the field corn.
Under the condition that the farming flow information corresponding to a certain corn field block accords with the farming flow characteristics of other corn types, the certain corn field block can be determined to be the corn field of other types.
According to the big data-based seed production corn field identification method, the difference of field corn and seed production corn in the farming flow is utilized, the target terminal signaling data used for representing farming flow information is introduced into the fine remote sensing classification of the corn field, and the supplement discrimination of the seed production corn is developed; the limitation of the remote sensing identification of the seed corn based on the optical characteristics can be overcome by introducing the agricultural process characteristics based on the traditional optical remote sensing characteristics, and the identification precision of the seed corn field based on the supplementary identification of the seed corn based on the agricultural process characteristics is beneficial to improvement.
In some embodiments, determining a corn field based on target sample point data within a target region and target remote sensing data corresponding to the target region comprises:
classifying the ground object types included in the target remote sensing data based on an unsupervised classification algorithm to obtain an optimal classification result of the target area;
and determining the corn field blocks based on the target sample point data and the optimal classification result.
The embodiment of the invention is mainly completed through the following three steps:
1. and (5) non-supervision classification of satellite remote sensing images.
In actual execution, based on an unsupervised algorithm, an image processing system is utilized to perform unsupervised classification on satellite remote sensing images in target remote sensing data. In the execution process, the input data of the image processing system are spectrum bands sensitive to vegetation types, such as red bands, red side bands, near infrared bands and the like, of medium-high resolution hyperspectral (or multispectral) satellite remote sensing data.
And carrying out a plurality of non-supervision classification experiments by adjusting the number of classification categories. And (3) carrying out superposition contrast analysis on each classification experiment result and the satellite remote sensing image by a visual interpretation method, and determining the optimal classification result of the target area. From the visual effect, most of the optimal classification results only comprise one type of ground objects, and meanwhile, most of the same ground objects are classified into one optimal classification result.
2. And (5) discriminating the type of the corn field.
And combining target sample point data and a visual interpretation method, analyzing and judging the optimal classification result one by one to judge the unsupervised classification category of the characteristic corn field, and determining the corn field blocks in the target area.
According to the big data-based seed production corn field identification method, the remote sensing extraction of the corn field is carried out by utilizing the non-supervision classifier instead of the supervision classifier through the fine spectrum detection advantage of the remote sensing data in the red-side wave band, so that the preconditions of the supervision classification effect on the number, quality and other aspects of training sample points are effectively avoided, the potential problems of low model robustness and the like caused by the factors of few training sample points, uneven distribution and the like in a research area are avoided, and the popularization of the big data-based seed production corn field identification method is improved.
In some embodiments, prior to determining the corn field based on the target spot data and the optimal classification result, further comprising:
carrying out vectorization processing on the binarization grid image corresponding to the target area to obtain an initial corn field range vector;
and superposing the corn field range vector and the high-resolution satellite remote sensing data in the target remote sensing data, and correcting the superposed data to obtain the target corn field range vector.
In actual implementation, binarization processing is carried out on the corn Tian Shange images after unsupervised classification, the classification result representing the corn field is assigned to 1, and other classification results are assigned to 0, so that a binarized grid image is obtained. The binarization process can greatly reduce the data amount in the raster image, thereby reducing the calculation amount and being capable of highlighting the outline of the corn field.
The binarized raster image is then vectorized to convert the corn field raster image to an initial corn field range vector, i.e., raster data is converted to a vector data structure representation.
And finally, superposing the initial corn field range vector on the high-resolution satellite remote sensing image, checking the corn field range vectors obtained by superposing the data one by one based on a visual interpretation method, and modifying the individual corn field range vectors with the shape and the size which are obviously inconsistent with the contours of the features corresponding to the images by manual vector editing to obtain a final range vector of the corn field block, namely a target corn field range vector.
It is understood that after determining that the corn fields included in the first region and the second region are seed corn fields, the first region and the second region may be used as target seed corn fields, that is, the seed corn field identification based on big data of the target region is completed, and then the target corn field range vector corresponding to the target seed corn field may be presented as the corn field identification result based on big data.
According to the big data-based corn field identification method, the corn field identification result based on big data is presented in a field vector mode, a discrete pixel-based result presentation mode of a direct classification identification method is replaced, the method is closer to the actual form of the seed corn field, and the readability is higher.
In some embodiments, determining whether each field is a seed corn field based on target terminal signaling data corresponding to each field in the second region includes:
extracting position information of a terminal and time information corresponding to the position information based on target terminal signaling data;
and determining at least one field as a seed corn field when the position information is at least one field in the second area and the time information is the ear period.
It can be understood that the agricultural processes such as female parent emasculation, water and fertilizer supplementing, male parent deforestation and the like different from the field corn can be carried out in the ear period of the seed corn field, so that the agricultural personnel can enter the seed corn field for many times in the ear period of the seed corn field, and the terminal geographic position track and the time track corresponding to the agricultural processes can be recorded through the target terminal signaling data. However, the agricultural processes of female parent emasculation, water and fertilizer supplementing, male parent deforestation and the like are not performed in the ear period of the field corn. Thus, the target terminal signaling data will not typically be present in the field corn at the ear stage.
In actual implementation, the target terminal signaling data may determine that at least one field in the second area is a seed corn field when the time information is determined to be the ear period and the position information is determined to be at least one field in the second area.
In the case that the second area includes a plurality of corn fields, target terminal signaling data of each corn field may be extracted, and time information and position information corresponding to each corn field may be determined.
When the position information is at least one corn field in the second region and the time information corresponding to the at least one corn field is a spike period, it is determined that the at least one field is a seed corn field.
It is understood that in the case where the position information is not at least one corn field in the second region or the time information corresponding to the at least one corn field is not a spike period, it may be determined that the seed corn field is not included in the second region.
According to the big data-based seed production corn field identification method, the information representing the farming flow can be extracted through the signaling data of the target terminal, and the big data-based seed production corn field identification precision can be improved.
In some embodiments, the time information includes: the total residence time of the spike period in a single day, the average residence time of the seeding period in a day and the average residence time of the growing period in a day;
the location information includes: the number of single-day terminals in the spike period, the number of average terminals in the seeding period and the number of average terminals in the growing period;
in the case that the position information is at least one field in the second area and the time information is the ear period, before determining that the at least one field is a seed corn field, the method further includes:
determining that the position information is at least one field in the second area and the time information is a spike period when the position information and the time information meet the target condition;
the target conditions include:
the single-day residence time of the spike period is within a time threshold range, and the time threshold range is determined based on the daily average residence time of the seeding period;
The number of the single-day terminals in the spike period is in a quantity threshold range, and the quantity threshold range is determined based on the number of the average terminals in the sowing period;
the daily residence time of the spike period is longer than the daily residence time of the growing period;
the number of single-day terminals in the spike period is larger than the number of average-day terminals in the growing period.
In actual implementation, firstly, statistics is performed on target terminal signaling data of a certain corn field in the second area in 3 periods of corn sowing period, growing period, spike period and the like, and time information and position information for representing a terminal space-time track are determined, which may include the following 6 parameters: the single-day residence time of the spike period, the single-day terminal number of the spike period, the average day residence time of the seeding period, the average day terminal number of the seeding period, the average day residence time of the growing period and the average day terminal number of the growing period.
The residence time represents the residence time of the terminal in a certain field and is used for representing time information, and the number of the terminals represents the number of times the terminal appears in a certain field and is used for representing position information.
It should be noted that the residence time of a single day is the sum of the residence time of all terminals in a certain corn field on a certain day; the number of terminals in a single day is the total number of terminals in a certain corn field; the average residence time of day is the sum of residence time of all terminals in a certain corn field block in a certain period divided by the total number of days; the average number of terminals per day is the total number of terminals that reside in a corn field for a period divided by the total number of days.
For example: the single-day residence time of the ear period is the sum of residence time of all terminals in a certain corn field in any day in the ear period; the number of the single-day terminals in the ear period is the total number of the terminals which reside in a certain corn field in any day in the ear period; the daily average residence time of a seeding period is the sum of residence time of all terminals which reside in a certain corn field block in the seeding period divided by the total number of days of the seeding period; the average number of terminals in the sowing period is the total number of terminals in a certain corn field in the sowing period divided by the total number of days in the sowing period; the average residence time in the growing period is the sum of residence times of all terminals in the growing period which reside in a certain corn field block divided by the total number of days in the growing period; the number of terminals on average in the growing period is the total number of terminals in the growing period that reside in a certain corn field divided by the total number of days in the growing period.
Based on target conditions, the fringe period single-day residence time is compared with the average residence time of the growing period, the fringe period single-day residence time is compared with the average residence time of the sowing period, the fringe period single-day terminal number is compared with the average terminal number of the growing period, and the fringe period single-day terminal number is compared with the average terminal number of the sowing period.
For the corn field for seed production, the residence time of the single day in the ear period and the number of the single day terminals in the ear period are respectively similar to the residence time of the average day in the sowing period and the number of the average day terminals in the sowing period, and a time threshold range and a number threshold range can be respectively set. The fringe single day residence time needs to be within a time threshold range and the fringe single day terminal number needs to be within a quantity threshold range.
The time threshold range and the number threshold range may be determined according to actual requirements, and are not specifically limited herein.
Meanwhile, the single-day residence time of the seed production corn field in the ear period is larger than the average residence time of the growing period, and the number of single-day terminals in the ear period is larger than the average number of terminals in the growing period.
For convenience of description, assuming that the fringe period single day residence time of a certain corn field block in the second region is a, the fringe period single day terminal number is b, the sowing period average day residence time is c, the sowing period average day terminal number is d, the growing period average day residence time is e, and the growing period average day terminal is f, the target conditions can be quantized to 4 conditions as follows:
(1)0.9c≤a≤1.1c;
(2)0.9d≤b≤1.1d;
(3)a>e;
(4)b>f。
comprehensively considering the influence of the most concentrated farming characteristics of farmers, the positioning errors of terminal signaling data and other factors, and determining a certain corn field block in the second area as a seed production corn field under the condition that the number of days for which the position information and the spatial information meet the target conditions exceeds a threshold of days.
For example: and identifying a certain corn field block with the time information and the space information in the ear period meeting the 4 conditions for more than 2 days as a seed production corn field.
It will be appreciated that one or more corn fields to be identified may be determined randomly as desired, or each corn field in the second region may be identified sequentially one by one, without limitation.
According to the big data-based seed production corn field identification method provided by the invention, the advantages of the position information and the time information can be recorded through the signaling data of the target terminal, the identification process of the seed production corn field is implemented by step implementation instead of direct classification identification by utilizing the differences of seed production corn and field corn in the aspects of planting modes, specific growth period farming processes and the like, and therefore, the quick seed production corn field can be realized.
In some embodiments, based on account opening information such as the attribution of the terminal number, the terminal user with the attribution not being the target area can be regarded as a person or a non-farmer person temporarily visiting the target area, and the terminal signaling data of the user is removed.
Based on the time flow information of the terminal signaling data, the terminal signaling data of the same terminal user is checked, the terminal signaling data of the terminal user which appears outside the target area in the target period is regarded as abnormal terminal signaling data according to the actual condition of the target area, and the abnormal terminal signaling data is removed.
For example: and regarding the signaling data of the mobile phone users crossing the surrounding administrative areas within 30 minutes as abnormal terminal signaling data, and eliminating the abnormal terminal signaling data.
Based on the corn field boundary and the time information included in the terminal signaling data, the terminal user with the time interval (i.e. residence time) smaller than the residence time threshold value between the time of entering the target area and the time of exiting the target area is regarded as a passthrough user or a non-farming staff, and the terminal signaling data of the user is removed. The residence time threshold may be dependent on the actual range situation of the target area.
For example: and regarding the user with the time interval between the time of entering the field and the time of exiting the field being less than 10 minutes as a passer-by user or a non-farming staff, and eliminating the terminal signaling data of the user.
According to the big data-based seed production corn field identification method provided by the invention, the abnormal value of the initial terminal signaling data is removed, so that the effectiveness of the data can be improved.
The embodiment of the invention can select a certain area in a certain province as a research area, and the research area is rapidly identified and extracted by the identification method for the seed production corn.
The general idea of the embodiment of the invention is as follows:
Firstly, distinguishing a corn field and other ground objects in a research area based on a high-resolution remote sensing image in a sentinel No. 2; then, subdividing the corn field based on a high-resolution remote sensing image of a high score No. 2 and a visual interpretation method by utilizing the difference of spectrum textures of vegetation canopy of the field corn and the seed corn in the ear period, and identifying the seed corn field; and finally, carrying out supplementary discrimination on the corn field blocks which cannot be discriminated by the high-resolution No. 2 image based on mobile phone signaling data by utilizing the difference of the agricultural process of the field corn and the seed production corn in the ear period, and obtaining the seed production corn field blocks in the research area.
The technical effects of the embodiment of the invention can be represented in the following three aspects:
1. the embodiment of the invention utilizes the advantages of the high-resolution multispectral satellite remote sensing data in the sentinel No. 2 that the multispectral satellite remote sensing data has a plurality of vegetation red-edge wave bands, realizes the rapid distinction of corn fields and other crops through an unsupervised classification algorithm, extracts corn field blocks, weakens the limitation of a visual interpretation method in large-area corn field block recognition, and simultaneously avoids the limitation that sufficient samples are needed for modeling based on supervised classification algorithms such as random forest machine learning and the like.
2. According to the embodiment of the invention, the advantage of the high-resolution multispectral satellite remote sensing image in the aspect of fine texture detection is utilized, and the crop canopy spectral texture characteristic enhancement mode is explored by means of the difference of planting modes of field corn and seed corn, namely, the seed corn field is identified based on the high-resolution multispectral characteristic by virtue of unique vegetation canopy spectral texture characteristics after female parent emasculation and male parent deforestation in the seed corn ear period stage, so that spectral similarity interference of the corn field in other growth periods during remote sensing fine classification is weakened.
3. According to the embodiment of the invention, based on the advantage that the spatial position can be recorded by the mobile phone signaling data, mobile phone signaling parameters reflecting the characteristics of the farming flow, such as the residence time of mobile phone users, the number of users and the like in the corn field blocks in the ear period are extracted, and the identification precision of the seed corns in a research area is improved by utilizing the farming flow difference of the seed corns and the field corns to carry out the supplement discrimination of the seed corns in the local corn fields with poor imaging quality of remote sensing images and insignificant high-resolution spectral characteristics.
Fig. 3 is a general technical flowchart of a method for identifying a corn field for seed production based on big data provided by the invention. Referring to fig. 3, the method for identifying the seed production corn field based on big data provided by the invention comprises the following steps:
The field investigation is carried out on the differences of the planting mode, the farming flow, the key nodes and the like of the seed corns in the research area and the field corns. In addition, through the field investigation method, field corn field sample point data and seed production corn field sample point data are collected and used for corn field remote sensing extraction and precision evaluation of seed production corn field identification based on big data.
Step 1.1, determining a planting mode
By a field investigation method, it is determined that the research area of the embodiment of the invention adopts a row ratio planting method, and the field planting row ratio of the parent plants and the female parent plants of the seed production corn field is set to be 1:7-1:10 inequality, namely 1 row of the parent plants are planted in the seed production corn field, and 7-10 rows of the female parent plants are correspondingly planted.
Step 1.2, determining the farming flow
And determining the farming flow of the corn field for seed production in the research area by a field investigation method. The difference between seed corn and field corn in the agronomic flow is mainly characterized in the spike period. In the ear period, the corn seeds need to be organized to enter the corn fields for farming for a plurality of times: firstly, after the seed production corn is subjected to emasculation, the stamens of a female parent plant needs to be removed in time, female parent emasculation is performed for short, and the selfing of the pistil and stamens of the female parent of the seed production corn is avoided; secondly, before and after pollination of the seed production corn, water and fertilizer are supplemented to the seed production corn; thirdly, after pollination of the seed production corn, the male parent plant needs to be cut down. For field corn, this is not generally done during the ear period.
The seed production corn field in the research area of the embodiment of the invention accords with the conventional farming flow of the seed production corn field.
Step 1.3, determining a time node
In the research area of the embodiment of the invention, corn starts sowing in the last ten days of 4 months, the period of 5-6 months is the corn growing period, the period of 7 months starts, and local corn successively enters the ear period.
Step 1.4, collecting sample points
The corn sampling points in the research area are collected by using a handheld global positioning system receiver through a field investigation method and are used for corn field remote sensing extraction and seed production corn field identification precision evaluation based on big data. The types of spots include field corn spots and seed corn spots. Various samples are required to be collected in standard and standard seed production corn fields, the samples are not required to be too aggregated, and the samples are required to be spatially scattered in the whole embodiment research area so as to ensure that the samples are typical and representative.
Step 2.1, data acquisition
And acquiring satellite remote sensing images of the ear period (after the male parent is cut down) of the seed production corn in the research area.
The remote sensing data adopted by the embodiment of the invention are sentencel number 2 satellite remote sensing data (Sentinel-2) and high-score number 2 satellite remote sensing data (GF-2). The sentinel No. 2 satellite is a medium-high resolution multispectral imaging satellite, has 13 spectral bands, has the spatial resolution of 4 bands such as red, green, blue and near infrared of 10 meters, and the sentinel No. 2 satellite is equipped with 3 vegetation red side bands in vegetation spectral red side region, and the spatial resolution is 10 meters, can acquire more careful vegetation spectral information. The high-resolution No. 2 satellite is a high-resolution multispectral imaging satellite and is provided with 2 high-resolution 1-meter full-color and 4-meter multispectral cameras. Fig. 4 is a schematic view of a high-resolution image No. 2 of the study according to the present invention, as shown in fig. 4.
And acquiring mobile phone signaling data of the research area.
The acquisition time of the mobile phone signaling data is respectively the corn sowing time, the growing period and the ear period of the research area. The mobile phone signaling data adopted by the embodiment of the invention is mobile phone signaling data provided by an operator.
The embodiment of the invention also obtains satellite remote sensing images with the spatial resolution of 1 meter in the research area through the geographic information platform, and the satellite remote sensing images are used as reference base images for multi-source image space registration. In addition, the relevant vector data such as the boundary of the research area is obtained and is used for cutting remote sensing satellite raster data and the like.
Step 2.2, data preprocessing
And carrying out preprocessing of multi-source remote sensing data, and respectively carrying out data preprocessing operations such as data fusion, orthographic correction, spatial registration, projection conversion, data clipping and the like on the multi-source remote sensing data.
The embodiment of the invention is carried out on remote sensing image processing platforms such as Pixel Information Expert, ERDAS, ENVI, arcGIS and the like, and the fused high-resolution No. 2 data is multispectral data with 0.8 meter spatial resolution.
Preprocessing of mobile phone signaling data is carried out, related parameters such as space coordinates and time nodes which can represent the user position flowing track in the signaling data are screened and mined, and vectorization processing is carried out on mobile phone signaling data parameters which represent the user position flowing track and the time track.
And carrying out preprocessing of sample data, and carrying out vectorization processing on the sample data acquired in the field based on the spatial position information. Through the data preprocessing step, the obtained sentinel No. 2 image, high-resolution No. 2 image, the extracted mobile phone signaling parameters, the acquired sample point data and other multi-source data are distributed into the same geographic reference system, for example, WGS84 coordinate system and Albers projection can be adopted.
Step 3, remote sensing extraction of corn field blocks
The embodiment of the invention utilizes the advantages of the sentinel number 2 data with a plurality of vegetation red-edge wave bands to roughly classify the ground objects in the image based on an unsupervised classification algorithm and corn sample point data acquired in the field so as to distinguish corn and other ground object types.
Step 3.1, unsupervised classification of images
The embodiment of the invention utilizes a ERDAS IMAGINE 2013 image processing system to carry out non-supervision classification on images. In the execution process, input data are blue wave band, green wave band, red wave band, near infrared wave band and3 vegetation red wave bands of the sentinel No. 2 satellite remote sensing data.
Key parameters for the unsupervised classification are set as follows: the loop convergence threshold Convergence Threshold =0.990, the maximum number of iterations Maximum Iterations =24, color scheme option Color Scheme Options (red=band 4, green=band 5, blue=band 3), other parameters remain default or default settings.
Based on the non-supervision classification parameter configuration, carrying out a plurality of non-supervision classification experiments by adjusting the classification category number. And carrying out superposition contrast analysis on each classification experiment result and the satellite remote sensing image, wherein a visual interpretation method shows that the non-supervision optimal classification result is obtained in the research area of the embodiment of the invention when the non-supervision classification category is set to 40, namely the classification result is 40 categories.
From the visual effect, most classification results only comprise one type of ground object, and meanwhile, the same ground object is classified into one classification result.
Step 3.2, discriminating the type of the corn field
Experiments show that in 40 types of non-supervision classification results, the classification results of the 3 types of 2, 8, 32 and the like correspond to corn fields in an original image, and the classification result of the corn fields in the research area provided by the invention is shown in figure 5.
Step 3.3, unsupervised post-Classification processing
Firstly, carrying out binarization processing on grid images subjected to unsupervised classification, and assigning a classification result representing a corn field to 1 and assigning other classification results to 0.
The binarized grid image is then vectorized to convert the corn Tian Leibie grid into a corn field range vector.
And finally, overlapping the corn field vector range with a high-resolution No. 2 image, checking the corn field range vectors one by one based on a visual interpretation method, and modifying individual corn field range vectors with the shape and the size which are obviously inconsistent with the profile of the corresponding land object of the image by manual vector editing to obtain the final range vector of the corn field.
Step 4, remote sensing identification of seed production corn fields
Based on the spectral texture difference of the corn field of seed production in the ear period and the corn layer of the field, based on a visual interpretation method, the identification of the corn field of seed production based on remote sensing data is realized. Female parent emasculation treatment and seed production corn male parent deforestation treatment are carried out in the seed production corn ear period, and the spectrum texture of the seed production corn male parent deforestation treatment shows stripe characteristics on a high-resolution No. 2 image. The spectrum texture of the field corn ear period is relatively homogeneous on the high-resolution No. 2 image because the female parent emasculation and the male parent deforestation are not carried out. Examples of canopy spectral textures of spike stage seed corn and field corn are shown in fig. 2.
The embodiment of the invention utilizes the agricultural processes of female parent emasculation, water and fertilizer supplementing, male parent deforestation and the like in the ear period of the seed production corn, which are different from field corn, and carries out secondary (supplement) identification on undetermined corn field blocks based on the characteristic that the geographic position of a mobile phone user can be recorded by mobile phone signaling data. The main basis is that in the ear period of the corn field, local farmers can enter the corn field for multiple times due to the agricultural processes of female parent emasculation, water and fertilizer supplementing, male parent deforestation and the like, and the geographical position track of the agricultural processes can be recorded through mobile phone signaling data. However, the field corn does not carry out the agricultural work processes of female parent emasculation, water and fertilizer supplementing, male parent cutting and the like in the ear period, so the mobile phone signaling data generally does not appear in the field corn field in the ear period.
Step 5.1, outlier rejection
And carrying out outlier rejection on the mobile phone signaling data to improve the effectiveness of the data.
Based on the account opening information of the resident and the like, the mobile phone user with the resident not being the research area is regarded as a person or a non-farmer visiting the research area temporarily, and the mobile phone user is rejected. Based on the time flow information of the signaling data of the mobile phone, the signaling data of the same mobile phone user is checked, and the mobile phone user with the cross-surrounding administrative area within 30 minutes is regarded as abnormal data and is rejected. Based on corn field boundary and mobile phone signaling data time information, a user with a time interval (residence time) between entering and exiting fields of less than 10 minutes is regarded as a passer-by user or a non-farming personnel, and is rejected.
Step 5.2, supplement discrimination
According to the embodiment of the invention, based on 2 variables such as residence time, number of users and the like, the corn fields which cannot be distinguished by the high-resolution No. 2 image texture characteristics are secondarily distinguished one by one.
Firstly, based on terminal signaling data, statistics of daily residence time and daily user number are carried out on the corn field in 3 periods such as corn sowing period, growing period and spike period, and further daily residence time and daily user number can be obtained.
And then, comparing the daily residence time of the corn field in the ear period and the daily user number of the corn field in the ear period with the daily residence time of the corn field in the seeding period, the daily user number of the corn field in the seeding period, the daily residence time of the corn field in the growing period and the daily user number of the corn field in the growing period respectively, and obtaining 4 results, namely, the residence time of the corn field in the ear period and the daily residence time of the corn field in the growing period, the residence time of the corn field in the ear period and the daily residence time of the corn field in the seeding period, the user number of the corn field in the ear period and the daily user number of the corn field in the growing period and the daily user number of the corn field in the ear period.
Finally, considering that personnel in agricultural links such as corn planting, emasculation of corn female parent, water and fertilizer supplementing, cutting of male parent and the like are usually relatively fixed for the same seed production corn field, therefore, for the seed production corn field, the residence time a of the ear period and the number b of the ear period should be similar to the average residence time c of the seeding period and the average number d of the seeding period, and simultaneously, the residence time a of the ear period and the number b of the ear period of the seed production corn field should be larger than the average residence time e of the growing period and the average number f of the growing period.
The embodiment of the invention quantifies the agronomic rule into the following discriminant rule, and the following 4 conditions should be theoretically satisfied for the corn field for seed production: (1) 0.9 c.ltoreq.a.ltoreq.1.1 c, (2) 0.9 d.ltoreq.b.ltoreq.1.1 d, (3) a > e and (4) b > f.
Comprehensively considering the influence of the factors such as the characteristics of most peasants with concentrated farming, positioning errors of mobile phone signaling data and the like, the embodiment of the invention judges the corn field blocks meeting the conditions for 2 days or more in the ear period as the seed production corn field.
Step 6, precision evaluation and result drawing
And (3) superposing a seed production corn field supplement judgment result based on mobile phone signaling data on the basis of extracting seed production corn field blocks by using remote sensing data, and forming a research area seed production corn field identification final result diagram based on big data. And evaluating the precision of the identification of the seed production corn field in the research area based on big data by utilizing a confusion matrix method through the sampling point data acquired in the seed production corn field.
In the embodiment of the invention, the overall classification precision is 93.88%, the precision of the seed corn producer is 96.04%, and the precision of the seed corn producer is 95.10%.
In order to further evaluate the effect of mobile phone signaling data on the identification of the seed corn field, the embodiment additionally evaluates the identification precision of the seed corn field when the mobile phone signaling data is not introduced (only based on remote sensing data), and the result shows that compared with the identification precision of the seed corn field only based on the remote sensing data, the overall classification precision, the user precision and the producer precision of the embodiment of the invention are respectively improved by 4.08%, 2.10% and 3.92%. And (3) drawing the corn field blocks for seed production in the research area of the embodiment based on the ArcGIS platform by adding three elements of a legend, a scale, a compass and the like. The identification result of the corn field for seed production in the research area of the embodiment of the invention is shown in fig. 6.
The embodiment of the invention integrates the advantages of the middle-high resolution hyperspectral satellite remote sensing data and the middle-high resolution multispectral satellite remote sensing data in the vegetation red edge wave band, the advantages of the high resolution satellite remote sensing data in the aspect of fine texture detection and the advantages of the mobile phone signaling data in the aspect of recordable space positions, utilizes the differences of the seed production corn and the field corn in the aspects of planting modes, specific growth period farming processes and the like, realizes the seed production corn field identification based on big data by implementing the steps rather than the direct classification identification mode.
From the result presentation mode, the invention presents the corn field identification result based on big data in a field vector mode, replaces the discrete pixel based result presentation mode of the direct classification identification method, is closer to the actual form of the seed corn field, and has stronger readability. In addition, the invention weakens the dependence on the effectiveness, the sample point quality and the spectrum characteristics of the classifier, and improves the applicability of the method.
The big data based seed production corn field identification device provided by the invention is described below, and the big data based seed production corn field identification device described below and the big data based seed production corn field identification method described above can be correspondingly referred to each other.
Fig. 7 is a schematic structural diagram of a device for identifying a corn field for seed production based on big data. Referring to fig. 7, an embodiment of the present invention provides a seed production corn field identification device based on big data, which may include: a first determination module 710, an identification module 720, and a judgment module 730.
A first determining module 710, configured to determine a corn field block based on target sample point data in a target area and target remote sensing data corresponding to the target area;
the identifying module 720 is configured to identify, from the corn field block, a first area and a second area based on spectral texture features corresponding to the corn field block, where the first area is a seed-producing corn field, and the second area is a corn field in which a planting type is not identified by the spectral texture features;
a judging module 730, configured to determine that each field is a seed corn field when the target terminal signaling data corresponding to each field in the second area meets a target condition;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field.
According to the large data-based seed production corn field identification device, the difference of field corn and seed production corn in the farming flow is utilized, the target terminal signaling data used for representing farming flow information is introduced into the fine remote sensing classification of the corn field, and the supplement discrimination of the seed production corn is developed; the limitation of the remote sensing identification of the seed corn based on the optical characteristics can be overcome by introducing the agricultural process characteristics based on the traditional optical remote sensing characteristics, and the identification precision of the seed corn field based on the supplementary identification of the seed corn based on the agricultural process characteristics is beneficial to improvement.
In some embodiments, the first determining module 710 is further configured to:
classifying the ground object types included in the target remote sensing data based on an unsupervised classification algorithm to obtain an optimal classification result of the target region;
and determining the corn field block based on the target sample point data and the optimal classification result.
In some embodiments, the big data based seed production corn field identification device further comprises:
the extraction module is used for extracting the position information of the terminal and the time information corresponding to the position information based on the signaling data of the target terminal;
and the second determining module is used for determining that the at least one field is a seed corn field when the position information is at least one field in the second area and the time information is a spike period.
In some embodiments, the time information includes: the total residence time of the spike period in a single day, the average residence time of the seeding period in a day and the average residence time of the growing period in a day;
the location information includes: the number of single-day terminals in the spike period, the number of average terminals in the seeding period and the number of average terminals in the growing period;
big data-based seed production corn field recognition device still includes:
A third determining module, configured to determine that the location information is at least one field in the second area and the time information is a spike period if the location information and the time information satisfy a target condition;
the target conditions include:
the spike period single day residence time is within a time threshold range, and the time threshold range is determined based on the sowing period daily average residence time;
the number of the single-day terminals in the spike period is in a quantity threshold range, and the quantity threshold range is determined based on the number of the daily terminals in the seeding period;
the total residence time of the spike period in a single day is longer than the residence time of the spike period in a common day in the growing period;
the number of the single-day terminals in the spike period is larger than the number of the average terminals in the growing period.
In some embodiments, the seed corn device further comprises:
the vector processing module is used for carrying out vectorization processing on the binarization grid image corresponding to the target area to obtain an initial corn field range vector;
and superposing the initial corn field range vector and the high-resolution satellite remote sensing data in the target remote sensing data, and correcting the superposed data to obtain the target corn field range vector.
In some embodiments, the seed corn device further comprises:
The acquisition module is used for respectively acquiring initial sample point data, initial remote sensing data and initial terminal signaling data in the target area;
the preprocessing module is used for carrying out vectorization processing on the initial sample point data and the initial terminal signaling data to obtain the target sample point data and the target terminal signaling data;
preprocessing the initial remote sensing data to obtain target remote sensing data;
wherein the preprocessing operation includes at least one of: data fusion, orthographic correction, spatial registration, projection conversion, and data cropping.
Fig. 8 illustrates a physical structure diagram of an electronic device, as shown in fig. 8, which may include: processor 810, communication interface (Communications Interface) 820, memory 830, and communication bus 840, wherein processor 810, communication interface 820, memory 830 accomplish communication with each other through communication bus 840. Processor 810 can invoke logic instructions in memory 830 to perform a big data based method for identifying a seed corn field, the method comprising:
determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
Identifying a first area and a second area from the corn field based on the spectral texture features corresponding to the corn field, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified by the spectral texture features;
judging whether each field is a seed production corn field or not based on target terminal signaling data corresponding to each field in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field.
Further, the logic instructions in the memory 830 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product including a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of executing the big data based method for identifying a corn field for seed production provided by the above methods, the method comprising:
determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
identifying a first area and a second area from the corn field based on the spectral texture features corresponding to the corn field, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified by the spectral texture features;
judging whether each field is a seed production corn field or not based on target terminal signaling data corresponding to each field in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the big data based method for identifying a seed corn field provided by the above methods, the method comprising:
Determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
identifying a first area and a second area from the corn field based on the spectral texture features corresponding to the corn field, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified by the spectral texture features;
judging whether each field is a seed production corn field or not based on target terminal signaling data corresponding to each field in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment of the invention. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (7)
1. The method for identifying the seed production corn field based on the big data is characterized by comprising the following steps:
determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
identifying a first area and a second area from the corn field based on the spectral texture features corresponding to the corn field, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified by the spectral texture features; the spectral texture features include a striped feature and a homogeneous feature, the first region exhibiting a striped feature, the second region exhibiting a feature other than the striped feature and the homogeneous feature;
judging whether each field is a seed production corn field or not based on target terminal signaling data corresponding to each field in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field;
before determining the corn field based on the target sampling point data in the target area and the target remote sensing data corresponding to the target area, the method further comprises the following steps:
respectively acquiring initial sample point data, initial remote sensing data and initial terminal signaling data in the target area; the initial remote sensing data are satellite remote sensing data of corn in the ear period in the target area;
Vectorizing the initial sample point data and the initial terminal signaling data to obtain the target sample point data and the target terminal signaling data, wherein the initial terminal signaling data is used for representing a terminal user position flow track;
preprocessing the initial remote sensing data to obtain target remote sensing data; the target remote sensing data comprise sentinel No. 2 satellite remote sensing data and high-score No. 2 data after data fusion; the sentinel No. 2 satellite remote sensing data has a blue wave band, a green wave band, a red wave band, a near infrared wave band and 3 vegetation red edge wave bands arranged in a vegetation spectrum red edge region; the spatial resolution of the sentinel No. 2 satellite remote sensing data is 10 meters; the high-resolution No. 2 data after data fusion is multispectral data with 0.8 meter spatial resolution;
the target sampling point data, the target remote sensing data and the target terminal signaling data are distributed into the same geographic reference system;
the determining the corn field block based on the target sample point data in the target area and the target remote sensing data corresponding to the target area comprises the following steps:
classifying the ground object types included in the sentinel No. 2 satellite remote sensing data in the target remote sensing data based on an unsupervised classification algorithm to obtain an optimal classification result of the target area;
Determining the corn field block based on the target sample point data and the optimal classification result;
the determining whether each field is a seed corn field based on the target terminal signaling data corresponding to each field in the second area includes:
extracting position information of a terminal and time information corresponding to the position information based on the target terminal signaling data;
the time information includes: the total residence time of the spike period in a single day, the average residence time of the seeding period in a day and the average residence time of the growing period in a day;
the location information includes: the number of single-day terminals in the spike period, the number of average terminals in the seeding period and the number of average terminals in the growing period;
and determining that the at least one field is a seed corn field before the position information is at least one field in the second area and the time information is a spike period, wherein the method further comprises:
determining that the position information is at least one field in the second area and the time information is a spike period when the position information and the time information meet a target condition;
the target conditions include:
the spike period single day residence time is within a time threshold range, and the time threshold range is determined based on the sowing period daily average residence time;
The number of the single-day terminals in the spike period is in a quantity threshold range, and the quantity threshold range is determined based on the number of the daily terminals in the seeding period;
the total residence time of the spike period in a single day is longer than the residence time of the spike period in a common day in the growing period;
the number of the single-day terminals in the spike period is larger than the number of the average terminals in the growing period;
setting the fringe period single day residence time of any corn field block in the second area as a, the fringe period single day terminal number as b, the sowing period average residence time as c, the sowing period average terminal number as d, the growing period average residence time as e and the growing period average terminal as f, wherein the target conditions are as follows:
0.9c≤a≤1.1c;0.9d≤b≤1.1d;a>e;b>f。
2. the method for identifying a corn field for seed production based on big data as defined in claim 1, wherein the determining whether each field is a corn field for seed production based on the target terminal signaling data corresponding to each field in the second area comprises:
and determining that the at least one field is a seed corn field when the position information is at least one field in the second area and the time information is a spike period.
3. The big data based seed production corn field identification method of claim 1, further comprising, prior to said determining said corn field pieces based on said target sample point data and said optimal classification result:
Carrying out vectorization processing on the binarization grid image corresponding to the target area to obtain an initial corn field range vector;
and superposing the initial corn field range vector and the high-resolution satellite remote sensing data in the target remote sensing data, and correcting the superposed data to obtain the target corn field range vector.
4. A big data based seed corn field identification method as defined in any of claims 1-3, wherein the preprocessing operation comprises at least one of: data fusion, orthographic correction, spatial registration, projection conversion, and data cropping.
5. Seed production corn field recognition device based on big data, characterized by comprising:
the first determining module is used for determining corn field blocks based on target sample point data in a target area and target remote sensing data corresponding to the target area;
the identification module is used for identifying a first area and a second area from the corn field block based on the spectral texture characteristics corresponding to the corn field block, wherein the first area is a seed production corn field, and the second area is a corn field with a planting type not identified through the spectral texture characteristics; the spectral texture features include a striped feature and a homogeneous feature, the first region exhibiting a striped feature, the second region exhibiting a feature other than the striped feature and the homogeneous feature;
The judging module is used for judging whether each field block is a seed production corn field or not based on the target terminal signaling data corresponding to each field block in the second area;
the target terminal signaling data is used for indicating the farming flow information corresponding to each field;
the apparatus further comprises:
the acquisition module is used for respectively acquiring initial sample point data, initial remote sensing data and initial terminal signaling data in the target area; the initial remote sensing data are satellite remote sensing data of corn in the ear period in the target area;
the preprocessing module is used for carrying out vectorization processing on the initial sample point data and the initial terminal signaling data to obtain the target sample point data and the target terminal signaling data, wherein the initial terminal signaling data is used for representing a terminal user position flow track;
preprocessing the initial remote sensing data to obtain target remote sensing data; the target remote sensing data comprise sentinel No. 2 satellite remote sensing data and high-score No. 2 data after data fusion; the sentinel No. 2 satellite remote sensing data has a blue wave band, a green wave band, a red wave band, a near infrared wave band and 3 vegetation red edge wave bands arranged in a vegetation spectrum red edge region; the spatial resolution of the sentinel No. 2 satellite remote sensing data is 10 meters; the high-resolution No. 2 data after data fusion is multispectral data with 0.8 meter spatial resolution;
The target sampling point data, the target remote sensing data and the target terminal signaling data are distributed into the same geographic reference system;
the first determining module is specifically configured to:
classifying the ground object types included in the sentinel No. 2 satellite remote sensing data in the target remote sensing data based on an unsupervised classification algorithm to obtain an optimal classification result of the target area;
determining the corn field block based on the target sample point data and the optimal classification result;
the judging module is specifically configured to:
extracting position information of a terminal and time information corresponding to the position information based on the target terminal signaling data;
the time information includes: the total residence time of the spike period in a single day, the average residence time of the seeding period in a day and the average residence time of the growing period in a day;
the location information includes: the number of single-day terminals in the spike period, the number of average terminals in the seeding period and the number of average terminals in the growing period;
and determining that the at least one field is a seed corn field before the position information is at least one field in the second area and the time information is a spike period, wherein the method further comprises:
determining that the position information is at least one field in the second area and the time information is a spike period when the position information and the time information meet a target condition;
The target conditions include:
the spike period single day residence time is within a time threshold range, and the time threshold range is determined based on the sowing period daily average residence time;
the number of the single-day terminals in the spike period is in a quantity threshold range, and the quantity threshold range is determined based on the number of the daily terminals in the seeding period;
the total residence time of the spike period in a single day is longer than the residence time of the spike period in a common day in the growing period;
the number of the single-day terminals in the spike period is larger than the number of the average terminals in the growing period;
setting the fringe period single day residence time of any corn field block in the second area as a, the fringe period single day terminal number as b, the sowing period average residence time as c, the sowing period average terminal number as d, the growing period average residence time as e and the growing period average terminal as f, wherein the target conditions are as follows:
0.9c≤a≤1.1c;0.9d≤b≤1.1d;a>e;b>f。
6. an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the big data based method of identification of a seed corn field as claimed in any one of claims 1 to 4 when the program is executed.
7. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the big data based seed corn field identification method of any of claims 1 to 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210407010.1A CN114743112B (en) | 2022-04-18 | 2022-04-18 | Big data-based seed production corn field identification method and device and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210407010.1A CN114743112B (en) | 2022-04-18 | 2022-04-18 | Big data-based seed production corn field identification method and device and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114743112A CN114743112A (en) | 2022-07-12 |
CN114743112B true CN114743112B (en) | 2023-05-12 |
Family
ID=82280590
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210407010.1A Active CN114743112B (en) | 2022-04-18 | 2022-04-18 | Big data-based seed production corn field identification method and device and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114743112B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111091052A (en) * | 2019-11-07 | 2020-05-01 | 中国农业大学 | Corn lodging area extraction system and method based on maximum likelihood method |
CN114240980A (en) * | 2021-11-08 | 2022-03-25 | 北京市农林科学院信息技术研究中心 | Unmanned aerial vehicle image sub-field boundary extraction method and device |
CN114241328A (en) * | 2021-11-08 | 2022-03-25 | 北京市农林科学院信息技术研究中心 | Method and device for extracting field and field straw based on aerospace remote sensing data fusion |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150206255A1 (en) * | 2011-05-13 | 2015-07-23 | HydroBio, Inc | Method and system to prescribe variable seeding density across a cultivated field using remotely sensed data |
CN106373150A (en) * | 2016-08-15 | 2017-02-01 | 中国农业大学 | Identification method of corns for seed on the basis of high resolution remote sensing data texture analysis |
CN106355143A (en) * | 2016-08-25 | 2017-01-25 | 中国农业大学 | Seed maize field identification method and system based on multi-source and multi-temporal high resolution remote sensing data |
CN107480706A (en) * | 2017-07-24 | 2017-12-15 | 中国农业大学 | A kind of seed production corn field remote sensing recognition method and device |
CN112395914A (en) * | 2019-08-15 | 2021-02-23 | 中国科学院遥感与数字地球研究所 | Method for identifying land parcel crops by fusing remote sensing image time sequence and textural features |
CN110751035A (en) * | 2019-09-16 | 2020-02-04 | 中国农业大学 | Seed corn production identification method and device |
CN114282726A (en) * | 2021-12-24 | 2022-04-05 | 中国电信股份有限公司 | Seed field prediction processing method and device, storage medium and electronic device |
-
2022
- 2022-04-18 CN CN202210407010.1A patent/CN114743112B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111091052A (en) * | 2019-11-07 | 2020-05-01 | 中国农业大学 | Corn lodging area extraction system and method based on maximum likelihood method |
CN114240980A (en) * | 2021-11-08 | 2022-03-25 | 北京市农林科学院信息技术研究中心 | Unmanned aerial vehicle image sub-field boundary extraction method and device |
CN114241328A (en) * | 2021-11-08 | 2022-03-25 | 北京市农林科学院信息技术研究中心 | Method and device for extracting field and field straw based on aerospace remote sensing data fusion |
Also Published As
Publication number | Publication date |
---|---|
CN114743112A (en) | 2022-07-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111709379B (en) | Remote sensing image-based hilly area citrus planting land plot monitoring method and system | |
Pang et al. | Improved crop row detection with deep neural network for early-season maize stand count in UAV imagery | |
CN104881865B (en) | Forest pest and disease monitoring method for early warning and its system based on unmanned plane graphical analysis | |
Culvenor | TIDA: an algorithm for the delineation of tree crowns in high spatial resolution remotely sensed imagery | |
CN112101159B (en) | Multi-temporal forest remote sensing image change monitoring method | |
US20100040260A1 (en) | Remote sensing and probabilistic sampling based method for determining the carbon dioxide volume of a forest | |
Sharma et al. | Geospatial technology in agroforestry: status, prospects, and constraints | |
CN113029971B (en) | Crop canopy nitrogen monitoring method and system | |
JP2010086276A (en) | Method and system for tree classification, method and system for generating current status information of forest, and method and system for selecting target area for tree thinning | |
CN112749627A (en) | Method and device for dynamically monitoring tobacco based on multi-source remote sensing image | |
CN108710864B (en) | Winter wheat remote sensing extraction method based on multi-dimensional identification and image noise reduction processing | |
CN114648705B (en) | Carbon sink monitoring system and method based on satellite remote sensing | |
CN113609899B (en) | Remote sensing time sequence analysis-based tilling land information positioning display system | |
CN116543316B (en) | Method for identifying turf in paddy field by utilizing multi-time-phase high-resolution satellite image | |
CN115272860A (en) | Method and system for determining rice planting area, electronic device and storage medium | |
Moradi et al. | Potential evaluation of visible-thermal UAV image fusion for individual tree detection based on convolutional neural network | |
CN114743112B (en) | Big data-based seed production corn field identification method and device and electronic equipment | |
Zhou et al. | A framework for precisely thinning planning in a managed pure Chinese fir forest based on UAV remote sensing | |
CN115471753A (en) | Tree species fine hierarchical classification method combining expert knowledge and machine learning | |
Dali et al. | Deep Learning Based Multi Task Road Extractor Model for Parcel Extraction and Crop Classification Using Knowledge Based Ndvi Time Series Data | |
do Amaral et al. | Geoprocessing Applied to Crop Management | |
Orny et al. | Characterization of maritime pine forest structure changes with vhr satellite imagery: application to the 24th january 2009 windfall damages cartography | |
Liu et al. | CropSight: Towards a large-scale operational framework for object-based crop type ground truth retrieval using street view and PlanetScope satellite imagery | |
Giovos et al. | Remote Sensing Vegetation Indices in Viticulture: A Critical Review. Agriculture 2021, 11, 457 | |
Álvarez | Monitoring Land Cover Changes using Landsat Data and Maximum Likelihood Classification: A Case Study of Hanoi, Vietnam |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |